Boris Cherny

Engineering Lead, Claude Code — Anthropic

Profile

Boris Cherny is the creator and Head of Claude Code at Anthropic — the AI coding agent that runs in your terminal and now writes roughly 4% of all public GitHub commits. He joined Anthropic in September 2024 and built Claude Code as a side project using an early Claude 3.6 model. Two months later a dogfood version shipped internally; by day five, half of Anthropic’s engineering team was using it. It launched publicly in February 2025 and crossed $1B in run-rate revenue within months, with Netflix, Uber, Spotify, Shopify, Ramp, Snowflake, Accenture, and Salesforce as paying customers.

Before Anthropic, Cherny spent five years at Meta, reaching Principal Engineer (IC8) — most of it on Instagram, where he led server architecture, development infrastructure, and large-scale codebase modernizations (Comet, Hack, GraphQL migrations). He has no CS degree: he studied economics, dropped out, founded his first startup at 18, and worked at a hedge fund before ending up in Big Tech. He’s the author of O’Reilly’s Programming TypeScript and maintains popular open-source projects like json-schema-to-typescript.

What makes Cherny interesting to developers isn’t just that he built Claude Code — it’s how he uses it. He runs five parallel Claude instances across separate checkouts, ships 20–30 PRs a day, and says he hasn’t personally written code in months. His design choices — CLAUDE.md files for project memory, hooks for automation, plan mode for iteration, MCP for tool integration — have effectively defined the grammar of AI-assisted development. If you’re using Claude Code to build anything (this website included), you’re working inside his mental model.

He’s also worth watching because he’s openly skeptical of where this ends. Cherny has said the interesting question isn’t how to be a better coder with AI, but what engineering looks like after coding is a solved problem. He thinks the bottleneck is shifting from code to taste, judgment, and product sense — a take that lands differently coming from the person whose tool is accelerating that shift.